Text in AI image

The simplest answer is to use AI to generate your image and overlay the text separately, in whatever design style you want to design it in. For most of your example images, that would be sufficient.

Unfortunately, most standard AI is notoriously bad at text. Some of the newer, more powerful models are much better.  I can't comment on the details of Storyline's AI assistant, but if you want better results, you may need to use external AI tools to create your vision. In-built tools are great for convenience, but often cannot compare to more dedicated tools for quality.

Here's a quick example I popped into a few different higher-end models, all first attempts.

Prompt:

word art graphic depicting the text "Come for the fritters, stay for the watermelons" centered on a southern-style rural background in 3-d lettering with yellow text on a green and brown background

Images:

Models:

Qwen Image        Flux 1.1 [Ultra]        Flux.1 Kontext [pro]

GPT Image 1         Flux.1 [dev]            Seedream 3.0